Biomarkers to Predict the Response to Cardiac Resynchronization Therapy

Ward Heggermont; Angelo Auricchio; Marc Vanderheyden


Europace. 2019;21(11):1609-1620. 

In This Article

Molecular Alterations in Cardiac Resynchronization Therapy Responders

In 2008, an important paper was published describing for the first time in detail some of the genetic alterations that occur in CRT responders.[34] Twenty-four patients underwent left ventricular biopsy procurement prior to CRT implantation, and 17 of them underwent biopsy procurement 4 months after implantation. Molecular markers at these time points were compared to a small control group of patients with normal cardiac function undergoing coronary artery bypass grafting. Responders to CRT were defined as having an increase in NYHA Class of >1, and a relative increase in LVEF of ≥25% at 4 months. Compared to the control group, HF patients had lower LV mRNA levels of alpha-myosin heavy chain (α-MHC), β-MCH, sarcoplasmic reticulum calcium ATPase 2-alpha (SERCA2α), phospholamban (PLN), and significantly higher BNP mRNA levels. The CRT responders had an increase in α-MHC, SERCA2α, an increased α-/β-MCH ratio, an increased SERCA2α/PLN ratio, and significantly lower BNP levels, while no significant changes in molecular profile were observed in non-responders. The results shed light on the genetic mechanisms inducing reverse remodelling, which is the underlying process resulting in a beneficial response to CRT.[35] However, these molecular markers cannot serve as useful biomarkers because procurement of LV biopsies is too invasive for this purpose.

In a large consecutive follow-up study, correlation between a CRT response and genetic variants at an allelic and genotypic level was performed.[36] A total of 207 patients out of 1421 were selected and divided between responders and non-responders, and were subsequently matched for their baseline parameters before CRT. A CRT response was defined as a decrease in LVESV of >15% at follow-up echocardiography compared to baseline. Genomic DNA, extracted from patient's blood during follow-up visit, was used to perform genotyping of a selection of genetic variants based on extensive literature search focusing on cardiovascular disease and vascular remodelling. Four genetic variants—both at allelic and genotypic level—were identified with CRT response, in the following genes: ATPIB1, guanine nucleotide-binding beta polypeptide 3 (GNB3), nuclear receptor subfamily 3 group C member 2 (NR3C2), and tumour necrosis factor superfamily member 11 (TNFSF11).[36] The further-reaching aspects of this study are however that machine learning algorithms were used to fine tune the prediction of a (non-)response, based on a combination of clinical parameters and the above-mentioned genetic variants.